industry 4.0 with intelligent manufacturing 5g mobile
TRANSCRIPT
Indonesian Journal of Electrical Engineering and Computer Science
Vol. 23, No. 3, September 2021, pp. 1376~1384
ISSN: 2502-4752, DOI: 10.11591/ijeecs.v23.i3.pp1376-1384 1376
Journal homepage: http://ijeecs.iaescore.com
Industry 4.0 with intelligent manufacturing 5G mobile robot
based on genetic algorithm
Mohd Zarifitri Kasim Hawari, Nur Ilyana Anwar Apandi Faculty of Electrical Engineering (FKE) Universiti Teknikal Malaysia Melaka (UTeM), Malaysia
Article Info ABSTRACT
Article history:
Received May 11, 2021
Revised Jun 4, 2021
Accepted July 4, 2021
A manufacturing fifth-generation (5G) mobile robot is a new development of
industry 4.0 application, deploying an unmanned system. This study aims to
implement a robot system for industrial applications in real-time with remote
sensors to enable humans. Moreover, there is still some obstacle to cope with
the better optimization solution for manufacturing 5G robot. This paper
proposed a latency network algorithm for the manufacturing 5G mobile robot.
An improved genetic algorithm (GA) by restructuring the genes is applied to
plan a mobile robot path. The process of the robot path in a complex workspace
is proposed, considering the node's collision-free constraint in the moving
phase of a robot. The proposed scheme improves the robot path and delivery
efficiency of the robot on average at 68% by moving on the industrial
environment's shortest path and time average of the mobile robot reach its
destination.
Keywords:
5G mobile robot
Genetic algorithm
Industry 4.0
Path planning
This is an open access article under the CC BY-SA license.
Corresponding Author:
Nur Ilyana Anwar Apandi
Faculty of Electrical Engineering (FKE)
Universiti Teknikal Malaysia Melaka (UTeM)
Jalan Hang Tuah Jaya, 76100 Durian Tunggal, Melaka, Malaysia
Email: [email protected]
1. INTRODUCTION
During the outbreak of the coronavirus pandemic, various companies and individuals involved in
developing the robotics industry have been working on various projects related to the pandemic [1], [2]. In any
project development, a robot must follow a path to safety, which is similar to the concept of industry 4.0.
According to [3], the need for high-performance and long-term data storage has become more prevalent in the
5G mobile network that has been developed. A 5G mobile robot can adapt to an industrial environment's
dynamic conditions that allow it to move seamlessly from one point to another.
On the other hand, an industrial internet of things (IIoT) is a new research area that focuses on
developing a modern system industrial ecosystem. Since the number of data sources is growing and the
heterogeneity of information collected, big data is defined not just by its volume but also by its
sophistication [4]. Thus, we propose the 5G mobile robot to suit the IIoT terms for the advanced technology
that will improve route planning efficiency and short time-consuming. Due to the complexity of the task and
the high cost of these components, 4G networks cannot handle the demand. Compared to previous technology,
5G data return technology is more reliable. The main advantages of 5G are faster processing speeds and lower
latency. The rise of the IoT technology has led to new processes and tools for factory automation. Since the
concept of the IoT has not been implemented in current automobile models, it is not feasible to implement a
real-time dynamic environment that can be used in self-driving cars, according to [5], [6].
Robotic path planning is a hot topic in robotics these days, particularly in robot navigation, which
involves determining a collision-free and optimal path for a mobile robot from beginning to end [7], [8]. This
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752
Industry 4.0 with intelligent manufacturing 5G mobile robot based on… (Mohd Zarifitri Kasim Hawari)
1377
topic focuses on the advantages of hybrid navigation in autonomous vehicles that demonstrated in [9]. While
these algorithms can solve the route planning problem, they all have drawbacks, such as the future field
method's local minimum problem. As a result, much work is still to be done to create a more effective
algorithm, according to [10], [11]. According to [12], the instability phenomena is intended. Due to the
instability phenomena, robots are prone to wandering away from their goal places, leading to catastrophic harm.
According to [13], a multi-robot that can exchange the sensing information data among the agents is proposed.
In [14], a Voronoi-based strategy is used to minimize duplicated exploration areas. This method works by
grouping different target locations into autonomous robots. According to [15], this paper proposes a multi-
robot system that coordinates with a wall follower of a group of robots. It is designed to evaluate the reliability
of the system. Mobile robot speed in a different scenario and speed control demonstrated in [16]-[18] is
proposed.
2. ROBOT PATH PLANNING BASED ON GENETIC ALGORITHM
The proposed genetic algorithm (GA) for path planning effectively optimizes a robot navigation
system's local and global planners, as demonstrated in [19]. The GA is a method of the quest for global
optimization, and it can model the evolution cycle and its behaviour in nature. A promising algorithm known
as the crossover operator in the GA has been implemented to increase the algorithm speed and accuracy, which
align with the works done in [20]-[24]. A population comprises the entire set of chromosomes and is estimated
based on the fitness function [25]-[26]. GA also implemented in network applications such as papers according
to [27] used the algorithm to find the shortest path for a network by using a combination of Open shortest path
first and genetic algorithm (OSGA). The algorithm can find a good set of values for the system's different
parameters used in visible light communication, as shown in [28]. In [29], GA with local search method is
proposed to uniform asymmetric multilevel inverter called USAMI. That removes the higher-order harmonics
defined while maintaining the fundamental voltage needed. The goal of the GA is to transform the primary
individuals into high-quality individuals. According to [30], GA has fast convergence towards global optimum
rather than A* algorithm for edge detection. The PSO algorithm has many shortcomings, such as its lack of
global convergence guarantee and trapped in its local optimum, as mentioned in [31]. Furthermore, some
academics are still looking into this topic to enhance the algorithm and find new approaches, and most of them
have solved the best path in terms of a single objective.
It is assumed for the user to an interface that will manufacture the manufacturing 5G mobile robot
based on the proposed route layout, which concentrating on the mobile robot's path that moves around inside
the designated area designed. We investigate the shortest path of the robot movement and the average time
taken by the robot. The optimization problem is considered by optimizing the network utility, and subsequently,
the appropriate algorithm is presented. We construct the model using a 2-D grid-based map to visualize our
environment for the mobile robot. Then, we apply the GA to the constructed model. The proposed GA is the
improvised from paper [21] that uses GA to rearrange the genes for the new population by testing three different
environments: irregular, maze and narrow winding environment. The GA was made according to the
improvement factors mentioned above in this study. Finally, we analyze the performance and average time of
the manufacturing 5G mobile robot based on the percentages differences of obstacles set. Unlike works in [21],
we leverage the algorithm from [22]-[25] by adding three different percentages of semi-dynamic obstacles.
Moreover, each set has six additional coordination consist of its existing obstruction.
This article is structured as follows; section 2 presents the manufacturing 5G mobile robot path
planning, and section 3 uses the GA approach for the mobile robot. In section 4, we present several results
obtained from the simulations. Finally, all the findings of the work are provided in section 5.
3. GENETIC ALGORITHM FOR 5G MOBILE ROBOT SCHEME
Through applying the GA, the steps to solve the problem are as follows: a chromosome coding system
was developed; initial population fitness; fitness feature assessment for initial population fitness; discovery,
crossover and mutation operator are chosen for the feasible solution. According to the theory of the fittest, the
highest value of expectation is achieved by the global search. The manufacturing 5G mobile robot route
planning intends to look for the shortest path across the working area. Route planning aims to guide the mobile
robot to find its suitable route to complete its movement from the starting point to the destination point.
Furthermore, using global route planning is used to build a collision-free path with restriction conditions. To
ensure the journey of the mobile robot from point to point is complete, specific data is given to the mobile
robot. Since the environment is a 2-D space, a grid-based space (where several obstacles are normal, known
and stable) is used in this proposed algorithm to represent the environmental space, and decimal encoding is
used according to [32]. In this article, the model's multi-layered representation is used to define the
ISSN: 2502-4752
Indonesian J Elec Eng & Comp Sci, Vol. 23, No. 3, September 2021: 1376 - 1384
1378
environmental space. Two-dimensional planar graphics represent the mobile robot's direction region, and a
circular list tracks the vertices (x,y) of the obstacle. The obstacle is set as an unspecified irregular polygon that
is static, random and identified.
The workspace as illustrates in Figure 1 is modeled as a grid with an irregular shape obstacle, where
each uniform cell points to a position. 𝑳 represent the length of the grid. The red line represents the path of the
robot to reach the destination. While symbol x represents the best possible vertices of the robot path. Thus, the
robot will find the optimal path to avoid the obstacle and safely arrive at the destination. Each area is split into
rows and columns that form the structure of a regular grid. Each cell must be rectangular but not necessarily
square in shape, the problem of complex environmental information is simple to solve and fewer resources are
required as mentioned in [33]. In the route planning of a mobile robot, the grid-based model is typically used
to show the workspace as it is simple to measure distances and represent obstacles. Let 𝑆𝑖 ( 𝑖 = 1, 2, 3, ⋯ , 𝑛)
represents as the area of obstacles covered/ shaded region where 𝑛 is the number of obstacles, 𝐿2 is the area of
the grid space. We assume to have ℎ the number of cells. The percentage of the shaded area which represents
the obstacle is calculated as below:
𝑆𝑖
𝐿2 × 100% = 𝑂ℎ% (1)
Ordered numbered grids define the entire workspace, and the number of grids defines how many cells
exist. Cells are defined by 𝐻, or 𝐻 ∈ [1: ℎ], where a total number of cells, ℎ = 20. Every cell is either
considered to be empty or occupied. Obstacles 𝑂ℎ represents the total percentages of the obstacles. These imply
that the cells occupied are off-limits for travel. Those obstacles are taken from the factory environment. The
obstacles boundary is established by their exact boundary plus minimum safety distance. In practice,
consideration of a mobile robot's size is 'one while 'moving. We assume that the percentage of the obstacle is
calculated with each sum of the cell. The shaded region represents all the obstacles in the form of irregular
shapes. Figure 2 illustrates the possible value for obstacle percentage and the allocation of number that defines
the value of obstacle percentage which (0,1). Step 1: the 𝑎 represent the full area of the obstacle which can be
defined as 1; Step 2: the 𝑏 and 𝑐 depict the small area of the obstacle area and non-covered area, respectively,
which is defined as 0. The path optimization is based on the GA method to find an optimal path for the
manufacturing 5G mobile robot. The optimization operator must analyse, choose, re-allocate, and perform
other optimization processes to improve individual fitness and optimise the GA convergence rate based on
individual gene characteristics and environmental factors that influencing individual genes.
Figure 1. Path planning with 2-d grid-based space
Figure 2. Possible value of each cell where 𝑎 = 1and 𝑏 = 𝑐 = 0
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752
Industry 4.0 with intelligent manufacturing 5G mobile robot based on… (Mohd Zarifitri Kasim Hawari)
1379
3.1. Genetic operator
3.1.1. Selection operator
To select the best gene selection and inherit the next generation by using the fitness function.
According to [26], the roulette method is used to ensure that the person who passes the obstacle is not selected.
3.1.2. Crossover operator
A single point crossover method is used to ensure that a person crosses by generating random
numbers. It produces an integer randomly within the range of genes as the intersection's starting point and
randomly generates an integer as an element to be crossed between populations. If the 𝑖-th individual
intersection is 𝑘 and the cross object is the 𝑗-th individual.
3.1.3. Mutation operator
Determine whether a randomly created number has mutated an individual. An integer within the range
of the number of genes is generated randomly as a variance point to ensure the diversity of the population
because of a minor change in chromosome position resulting in the formation of a new gene fragment.
Algorithm 1 is the whole process of the manufacturing 5G mobile robot path planning. As the output of GA is
𝑲 that denoted as the best fitness values of the coordination 𝑥 and 𝑦.
3.2. Chromosome coding
In the encoding of chromosomes, true coding, binary coding, and tree coding are used, which will
affect the usefulness of route planning. Here, the real number encoding method is used to express the
coordinates of real points, while the path coordinates are encoded specifically. The robot's direction consists
of multiple line segments, each represented by a line attached to the infinite starting point and endpoint node.
We represent the symbol £ as a collection of intermediate nodes. Assume that there are 𝑛 nodes overall, a point
𝑝𝑖 ( 𝑖 = 1, 2, 3, ⋯ , 𝑛) represents the 𝑖-th on the chromosome, that is £= {𝑝1,𝑝2, ⋯ , 𝑝𝑛}. According to [25], in
comparison to binary coding, there are several benefits of real coding: (1) Participate in the genetic process
where actual numbers are used as chromosomes, instead of the time-consuming method of coding and
decoding; (2) Removing the binary code 'hamming cliff issue'; (3) crossover and mutation operators are used
for breeding new individuals, which enhances the algorithm's search efficiency. These points are then
processed in a variety of ways to determine the best robot motion direction. Let 𝒍𝑪 implied as total number of
chromosome length.
Algorithm 1: The Main Function of GA Algorithm 2: Population Initialization
(𝒙, 𝒚)
Algorithm 3: Selection Operator for Gene
Selection using Fitness Function Value
Input: Environment setup
Output: Optimal Path for
Mobile Robot
1: Initialize: Crossover,
Mutation, Population Size
chromosome length, 𝒙 and 𝒚 limit)
2: for 𝒊 =1: 1: 𝑵 3: do function in 3.1.1;
3.1.2; 3.1.3
4: if rand <0.0
5: [𝒙, 𝒚]=optimization (𝒙, 𝒚) Function 6: end if
7: Calculate fitness
Function (best 𝒙, 𝒚) 8: end for
9: for 𝒊 = 1: 1: max (size
(best 𝒙, 𝒚) – 1 10: do (2) 11: end for
12: 𝑲 = plot best 𝒙, 𝒚
Input: Population number of
genes
Output: Initial population
1: Initialize: 𝑵𝒑, 𝒍𝑪,
𝒙𝟎, 𝒚𝟎, 𝒙𝒏, 𝒚𝒏
2: 𝒊= 1; for 𝒋=1:1: 𝑵𝒑
3: while 𝒊<= 𝒍𝑪
4: if 𝒊==1
5: 𝒙(𝒋, 𝒊) = 𝒙𝟎; 𝒚(𝒋, 𝒊) = 𝒚𝟎; 𝒊 = 𝒊 +
𝟏 6: else if 𝒊< 𝒍𝑪-1
7: if rand<0.5
8: else (2)
9: else if 𝒊== 𝒍𝑪
10: 𝒙(𝒋, 𝒊) = 𝒙𝒏; 𝒚(𝒋, 𝒊) = 𝒚𝒏; 𝒊 =
𝒊 + 𝟏; 11: end if
12: end while
13: 𝒊= 1 14: end for
Input: Initial population
adaptability
Output: New population (new
initial)
1: Initialize: new population
for 𝒙, 𝒚 = selection
(𝒙, 𝒚,fitness value)
2: 𝑭 = ∑ 𝑭𝒗𝟏𝟎𝒗=𝟏
3: 𝑭𝒗=∑ 𝑭𝒗𝒊𝟏𝟎𝒊=𝟏
4: while 𝑵𝒑<=𝑷𝒙
5: if 𝑭𝒗𝒊> 0
6: else if
7: 𝑭𝒗𝒊 = 𝑭𝒗𝒊 + 𝟏; 8: end if 9: end while
3.3. Initialize population
The population must be initialized first to plan for the optimum direction for the mobile robot. Since
the initial path is generated arbitrarily, the continuous domain contains the feasible path and the unreachable
path. The genetic operator would inevitably find a direction that is close to ideal. Due to the unreachable
direction, the utility search in the global search is reduced because where the chromosomes overlap,
unreachable paths are likely to be generated by the two unreachable paths. Therefore, in the road's generation,
this paper set an initial population to speed up the convergence rate to maintain the escape of obstacles and
ISSN: 2502-4752
Indonesian J Elec Eng & Comp Sci, Vol. 23, No. 3, September 2021: 1376 - 1384
1380
deselect the path that encountered obstacles. A single production point is used for the initialization process,
setting (𝑥0, 𝑦0) as the starting point, (𝑥𝑛, 𝑦𝑛) as the destination point, and meet 𝑥𝑛 ≥ 𝑥0, 𝑦𝑛 ≥ 𝑦0. The current
position is 𝑖, while the 𝑖 + 1 point is produced, where 𝑅 is a random function, [0,1]. The regulatory algorithm
for the initial population is shown in Algorithm 2. Let 𝑵𝒑 denoted as total number of new populations while
𝑷𝒙 represents as number of matrices from small to large for the number of populations.
3.4. Fitness function
The fitness function is the assessment of the individual's environmental adaptability. Based on Figure
3, the fitness function is produced through the process of the population. There is a lower fitness of the
individual who crosses the obstacle after the mutation and crossover. So, the person's fitness is set to 0 by the
obstacle, and the fitness of the other individual is stated. As a result, there are two types of conditions for the
adaptive operator: if the individual gets past the obstacles, it is possible to obtain fitness; while if the obstacles
cannot pass, then compute the total number of fitness values given by:
𝐹 = {0 if 𝑂ℎ is 0,
𝐶
∑ √(𝑥𝑖+1−𝑥𝑖)2+(𝑦𝑖+1−𝑦𝑖)2𝑛−1𝑖=1
if 𝑂ℎ is 1. (2)
Individual adaptability and fitness value are higher when 𝐶 is any constant. More elevated fitness
chromosomes are passed to the next generation, and if a healthy individual chromosome is found, this process
continues. From (2), 𝐹 denotes as best fitness for each population while 𝐹𝑣𝑖 is the total initial fitness values.
Algorithm 3 shows the step for fitness and gene selection. In summary, the flow process of GA for mobile
robot path planning is shown in Figure 3. The flowchart illustrates the overall GA optimization part used in
this research.
Figure 3. Flow chart of GA for mobile robot planning task
4. RESULTS AND DISCUSSION
To verify the effectiveness of the proposed algorithms, a simulation in one fair environment with
different percentages of the obstacles: 50%, 70% and 90%. The shaded part of the initial configuration is made
up of a coordination framework and an obstacle field. Table 1 shows the setting parameters of improved GA
with obstacles implemented. The values are set considering the actual environment of the industry path. The
fitness value is based on the GA parameters given to investigate the performance of fitness values. This value
minimizes the GA process and optimizes and finds the robot's optimal path to reach the destination. The selection
rate for the improved GA is 0 because the GA no need to choose a large number of chromosomes to be selected
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752
Industry 4.0 with intelligent manufacturing 5G mobile robot based on… (Mohd Zarifitri Kasim Hawari)
1381
and the process will take a shorter time to optimize. This research conducts three experiments which are using
three different numbers of percentages. We proposed six environments for each experiment. For each
experiment, the environment is provided with a possible value of obstacle coordination which means the obstacle
is semi-dynamic for experiments 1, 2 and 3. We randomly move the position of the obstacle with the aid of the
obstacle number provided. While Figure 4 shows the six possible environments of the obstacles for each
experiment; Figure 4(a) 50% obstacle; Figure 4(b) 70% obstacle; and Figure 4(c) 90% obstacle.
(a)
(b)
(c)
Figure 4. Proposed experiment for the various obstacle; (a) 50%; (b) 70%; (c) 90%
Table 1. GA parameters setting Parameters Proposed Values Parameters Proposed Values
Population Size, 𝑵𝒑 100 Selection Rate, 𝑷𝒔 0
Number of Genes 50 Crossover Rate, 𝑷𝒄 0.6
Maximum Number of Generation 100 Mutation Rate, 𝑷𝒎 0.01
The blocks are considered obstacles where the blocks' condition is to be semi-dynamic, which
indicates that an obstacle is present in each region where the robot travels from its starting point to its
destination point. The mobile robot starting position point is at (0,0), and the destination point is at (20,20). A
minimum and maximum value of the path length is shown to optimise the result obtained from these three
experiments. The robot path and the CDF value of the three minimum path lengths are illustrated in Figure
5(a)-(c) and Figure 6, respectively. The robot path represents in a red dotted line while the blue dots are the
fitness value of the coordination to avoid the obstacle with GA operation involvement. Hence, the GA will
optimize the path where the mobile robot is near the obstacle and make sure the path is smooth and shorter for
the mobile robot to reach its destination.
ISSN: 2502-4752
Indonesian J Elec Eng & Comp Sci, Vol. 23, No. 3, September 2021: 1376 - 1384
1382
(a)
(b)
(c)
Figure 5. Mobile robot path with various percentages of obstacle; (a) 50%; (b) 70%; (c) 90%
Figure 6 (a) shows that the minimum path length for 50%, 70%, and 90% of obstacles are 28.32, 29.91
and 35.33 meters, respectively. As we know, the more complex the environment in the area, the longer time
for the robot to get through the semi-dynamic obstacle. The robot path appears to be more consistent during
the 50% obstacle parameter was made, where the others are erratic for the mobile robot path planning until the
end of the 100 run times simulations. Based on Figure 6 (b), the mobile robot's maximum path length for 50%,
70%, and 90% of obstacles are 30.52, 32.06 and 35.58 meters, respectively. The robot path takes maximum
length in all three experiments. As our observation, when simulations ended at 100, the robot path's length for
maximum percentage implemented is recorded less value than the minimum set. The maximum percentage
only recorded 45 meters while 50 meters for the minimum. The robot path can be the same as the simulations
run until 100 times with the best individual genes and better fitness function production for each population.
(a) (b)
Figure 6. CDF; (a) minimum and (b) maximum total length for the various percentage of obstacle
In line with [19]–[20], the average speed for manufacturing a mobile robot is 0.5 with a maximum load of
1 kilogram. This setting is due to some limitations of the robot speed due to the safety in the industrial environment
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752
Industry 4.0 with intelligent manufacturing 5G mobile robot based on… (Mohd Zarifitri Kasim Hawari)
1383
and robot max payload. Table 2 shows the time average of the mobile robot with constant speed from start to end
position based on the percentage of obstacles. The design of the mobile robot is considered for stability and consistent
speed. Thus, the mobile robot can reach its destination on time with the presence of obstacles.
Table 2. Average time for various percentage of obstacle Percentage of obstacle (%) 50 70 90
Minimum; Maximum total length time (sec) 60.498; 61.859 66.792; 67.964 79.585; 79.985
Moreover, the GA application functioned as a selection process that makes it possible to select good-
quality solutions for mating in each generation and crossover activity that enables exchanging genetic
information between two good-quality solutions to create better solutions generation. Crossover plays an
important role in generating different gene combinations and has both potentials to explore and manipulate,
thus improving the solution's quality. Our path optimization operator can be seen from the simulations, no
matter how complicated or the higher the obstacles percentage of the environment is to shorten the path and
reduce the length to improve the robot's protection. The convergence speed of the GA operator is improving
the simulations before and after optimizations. As a result, the manufacturing 5G mobile robot gains the desired
path length in a shorter distance. It reflects this study, which aims at the path optimization operator, which can
speed up evolution and shorten path length in various environments using GA. While this is useful in improving
the objective value, the search is probably pointless if the current solution is stuck in the optimum local area.
5. CONCLUSION
In this paper, industry 4.0 with intelligent manufacturing 5G, a mobile robot based on a GA, is
proposed to optimize the mobile robot's path or route. According to the optimal path planning with different
percentages of obstacles covered without collision, this improved GA is optimized to reduce the path length
before and after optimization. The results of the experiments show that this strategy strengthens the ideal path
even further. As the outcome suggests, path planning for manufacturing 5G mobile robots based on improved
GA has developed better optimization than the previous GA. Future work improvement is proposed to develop
a further GA-based scheme in terms of performance, such as energy consumption with 5G implementation and
multi-robot involvements.
ACKNOWLEDGEMENTS
This work was supported by the Ministry of Higher Education Malaysia through the Fundamental
Research Grant Scheme (FRGS) by Universiti Teknikal Malaysia Melaka (UTeM) under Grant
FRGS/2018/FKE CERIA/F00355.
REFERENCES [1] Z. Song, et al., “From SARS to MERS, thrusting coronaviruses into the spotlight,” Viruses, vol. 11, no. 1, pp. 59,
2019, doi: 10.3390/v11010059.
[2] E. Guizzo, “Coronavirus pandemic: A call to action for the robotics community,” IEEE Spectrum, 2020.
[3] M. Agiwal, A. Roy, and N. Saxena, “Next Generation 5G Wireless Networks : A Comprehensive Survey,” IEEE Comm.
Surveys & Tutorials, vol. 18, no. 3, pp. 1617-1655, Feb. 2016, doi:10.1109/COMST.2016.2532458.
[4] A. Mosteiro-Sanchez, M. Barcelo, J. Astorga, and A. Urbieta, “Securing IIoT using Defence-in-Depth: Towards an
End-to-End secure Industry 4.0,” Journal of Manufacturing Systems, vol. 57, pp. 367-378, Oct. 2020,
doi:10.1016/j.jmsy.2020.10.011.
[5] K. S. Bhandari and G. I. H. Cho, “An Energy Efficient Routing Approach for Cloud-Assisted Green Industrial IoT
Networks,” Sustainability, vol. 12, no. 18, pp. 7358, Jan. 2020, doi:10.3390/su12187358.
[6] F. Al-Turjman and S. Alturjman, “5G/IoT-enabled UAVs for multimedia delivery in industry-oriented applications,”
Multimedia Tools and Applications, vol. 79, no. 13-14, pp. 8627-8648, Apr. 2018, doi: 10.1007/s11042-018-6288-7.
[7] Y. Wu, S. Wu, and X. Hu, “Cooperative Path Planning of UAVs & UGVs for a Persistent Surveillance Task in
Urban Environments,” IEEE Internet of Things Journal, vol. 8, no. 6, pp. 4906-4919, Oct. 2021,
doi:10.1109/JIOT.2020.3030240.
[8] J. Ni, W. Wu, J. Shen, and X. Fan, “An improved VFF approach for robot path planning in unknown and dynamic
environments,” Mathematical Problems in Engineering, Jan. 2014, doi:10.1155/2014/461237.
[9] J. Ni, K. Wang, H. Huang, L. Wu, and C. Luo, “Robot path planning based on an improved genetic algorithm with
variable length chromosome,” in 12th Int. Conf. on Natural Computation, Fuzzy Systems and Knowledge Discovery,
pp. 145-149. IEEE. Aug. 2017, doi:10.1109/FSKD.2016.7603165.
ISSN: 2502-4752
Indonesian J Elec Eng & Comp Sci, Vol. 23, No. 3, September 2021: 1376 - 1384
1384
[10] M. S. Gharajeh and H. B. Jond, “Hybrid Global Positioning System-Adaptive Neuro-Fuzzy Inference System based
autonomous mobile robot navigation,” Robotics and Autonomous Systems, vol. 134, pp. 103669, Oct. 2020,
doi:10.1016/j.robot.2020.103669.
[11] J. Ni, L. Wu, X. Fan and S. X. Yang, “Bioinspired intelligent algorithm and its applications for mobile robot control:
A survey,” Computational Intelligence & Neuroscience, vol. 2016, no. 2, pp. 1-16, Jan. 2016,
doi:10.1155/2016/3810903.
[12] Z. Chen, J. Zhou, R. Sun, L. Kang, “A new evolving mechanism of genetic algorithm for multi-constraint intelligent
camera path planning,” Soft Computing, vol. 25, no. 7, pp. 5073-5092, Jan. 2021, doi:10.1007/s00500-020-05510-6.
[13] M. H. Mohamad Nor, Z. H. Ismail, and M. A. Ahmad, “Broadcast control of multi-robot systems with norm-limited
update vector,” Int. Journal of Advanced Robotic Systems, vol. 17, no. 4, 2020, doi:10.1177%2F1729881420945958.
[14] E. Olcay, F. Schuhmann, and B. Lohmann, “Collective navigation of a multi-robot system in an unknown
environment,” Robotics and Autonomous Systems, vol. 132, p. 103604, 2020, doi: 10.1016/j.robot.2020.103604.
[15] J. Hu, H. Niu, J. Carrasco, B. Lennox, and F. Arvin, “Voronoi-Based Multi-Robot Autonomous Exploration in
Unknown Environments via Deep Reinforcement Learning,” IEEE Trans. on Vehicular Technology, vol. 69, no. 12,
pp. 14413-14423, Dec. 2020, doi:10.1109/TVT.2020.3034800.
[16] A. N. Jati, R. E. Saputra, M. G. Nurcahyadi, and N. T. Al Ghifary, “A Multi-robot System Coordination Design and
Analysis on Wall Follower Robot Group,” International Journal of Electrical and Computer Engineering (IJECE),
vol. 8, no. 6, p. 5098, Dec. 2018, doi:10.11591/ijece.v8i6.pp5098-5106.
[17] J.-G. Li, J. Yang, S. Cui, and L. Geng, “Speed limitation of a mobile robot and methodology of tracing odor plume
in airflow environments,” Procedia Eng., vol. 15, pp. 1041-1045, Jan. 2011, doi:10.1016/j.proeng.2011.08.192.
[18] V. N. Thanh, D. P. Vinh, N. T. Nghi, L. H. Nam, and D. L. H. Toan, “Restaurant Serving Robot with Double Line
Sensors Following Approach,” 2019 IEEE International Conference on Mechatronics and Automation (ICMA), 2019, doi:10.1109/ICMA.2019.8816404.
[19] L. Chang, L. S. C. Jiang, and Y. Dai, “Reinforcement based mobile robot path planning with improved dynamic
window approach in unknown environment,” Autonomous Robots, vol. 45, no. 1, pp. 51-76, Sep. 2020,
doi:10.1007/s10514-020-09947-4.
[20] M. Li, C. Wang, Z. Chen, X. Lu, M. Wu, and P. Hou, “Path planning of mobile robot based on genetic algorithm and
gene rearrangement,” 2017 Chinese Automation Congress (CAC), Oct. 2017, doi:10.1109/CAC.2017.8244039.
[21] C. Chen, Z. Shuanggui, Y. Qingying, Y. Zitong, Y. Zhen, and H. Fan, “Personalized travel route recommendation
algorithm based on improved genetic algorithm," Journal of Intelligent & Fuzzy Systems, vol. 40, no. 3,
pp. 4407-4423, Mac. 2021, doi:10.3233/JIFS-201218.
[22] M. A. Albadr, S. Tiun, M. Ayob, and F. T. Al-Dhief, “Genetic Algorithm Based on Natural Selection Theory for
Optimization Problems,” Symmetry, vol. 12, no. 11, p. 1758, Oct. 2020, doi:10.3390/sym12111758.
[23] L. M. Schmitt, “Theory of genetic algorithms,” Theoretical Computer Science, vol. 259, no. 1-2, pp. 1-61, May 2001,
doi:10.1016/S0304-3975(00)00406-0.
[24] Y. Wang, H. Zhou, and Y. Wang, “Mobile robot dynamic path planning based on improved genetic algorithm,”
Wireless Communications and Mobile Computing, Aug. 2017, doi: 10.1063/1.4992863.
[25] X. Gong, D. Plets, E. Tanghe, T. D. Pessemier, L. Martens, and W. Joseph, “An efficient genetic algorithm for large-
scale transmit power control of dense and robust wireless networks in harsh industrial environments,” Applied Soft
Computing, vol. 65, pp. 243-259, Apr. 2018, doi:10.1016/j.asoc.2018.01.016.
[26] T. Yifei, Z. Meng, L. Jingwei, L. Dongbo, and W. Yulin, “Research on Intelligent Welding Robot Path Optimization
Based on GA and PSO Algorithms,” IEEE Access, vol. 6, pp. 65397-65404, 2018, doi:10.1109/ACCESS.2018.2878615.
[27] T. Kamalakis, L. Dogkas, and F. Simou, “Optimization of a discrete multi-tone visible light communication system
using a mixed-integer genetic algorithm,” Optics Communications, vol. 485, pp. 126741, Apr. 2021,
doi:10.1016/j.optcom.2020.126741.
[28] H. C. W. Lau, T. M. Chan, W. T. Tsui, and W. K. Pang, “Application of Genetic Algorithms to Solve the Multidepot
Vehicle Routing Problem,” IEEE Trans. on Automation Science and Engineering, vol. 7, no. 2, pp. 383-392, Apr.
2015, doi:10.1109/TASE.2009.2019265.
[29] R. F. Abdelgoui, R. Taleb, A. Bentaallah, and F. Chabni, “Improved hybrid algorithm based on GA and local search
method for asymmetrical 9-level inverter,” Indonesian Journal of Electrical Engineering and Computer Science
(IJEECS), vol. 21, no. 3, p. 1309, Mar. 2021, doi:10.11591/ijeecs.v21.i3.pp1309-1316.
[30] R. Al-Jarrah, M. Al-Jarrah, and H. Roth, “A Novel Edge Detection Algorithm for Mobile Robot Path Planning,”
Journal of Robotics, vol. 2018, pp. 1-12, Jan. 2018, doi:10.1155/2018/1969834.
[31] G. Li and W. Chou, “Path planning for mobile robot using self-adaptive learning particle swarm optimization,”
Science China Information Sciences, vol. 61, no. 5, Nov. 2017, doi:10.1007/s11432-016-9115-2.
[32] J. Ma, Y. Liu, S. Zang, and L. Wang, “Robot path planning based on genetic algorithm fused with continuous bezier
optimization,” Computational intelligence and neuroscience, Feb. 2020, doi:10.1155/2020/9813040.
[33] X. Cao, C. Sun, and M. Chen, “Path planning for autonomous underwater vehicle in time-varying current,” IET
Intelligent Transport Systems, vol. 13, no. 8, pp. 1265-1271, Apr. 2019, doi:10.1049/iet-its.2018.5388.